The utilization of CEUS and SWE for predicting pathological complete response to neoadjuvant chemotherapy for invasive breast cancer.

Introduction
Breast cancer is the most common malignancy among women worldwide, and it is the main cause of cancer-related death in 112 countries1. Neoadjuvant chemotherapy (NAC) has been widely adopted for treating invasive breast cancer patients whose axillary or peripheral lymph nodes are involved to reduce the tumor burden2. Even for the same molecular subtypes of invasive breast cancer, some patients may respond well to NAC, but other patients may resist it. Pathological results are the gold standard for evaluating NAC. Pathological complete response (pCR) is associated with longer disease-free survival (DFS) and better overall survival (OS)3. In light of the late assessment of pathological treatment response, it is critical to combine effective early evaluation with personalized cancer remedies based on risk categories for the NAC regimen success4–6. There is an urgent need to develop more accurate imaging modalities for determining the degree of residual tumor after NAC and to provide information for selecting the most appropriate surgical plan and predicting the prognosis of patients.
Multiparametric ultrasound imaging approaches, including US, contrast-enhanced ultrasound (CEUS), and shear wave elastography (SWE), can be used to monitor the NAC response. They can display changes in the size, shape, microvascular perfusion and stiffness of the breast tumor during NAC. US is a low-cost, nonirradiating, noninvasive and safe examination method for assessing the NAC response based on the changes from pretreatment baseline imaging to the residual lesion after treatment7. It can be used as a primary screen for abnormal NAC responders8. Although tumor size is a relevant predictive factor of response to NAC, the changes in maximum tumor diameter by conventional US before and after NAC are far from sufficient for evaluating the efficacy of NAC9. It is necessary to develop more non-invasive ultrasonic imaging methods for improving prediction of pCR before surgery. CEUS, on the other hand, can be used to visualize the breast lesions through real-time tracking of microbubbles entering blood vessels, enable quantitative measurement of vascularity, and more accurately reveal the true size of the tumor for evaluating its NAC’s efficacy10. Previous studies have investigated the sensitivity and effectiveness of chemotherapy drugs by observing the changes in the tumor size and pathology of patients with breast cancer during NAC11. Patients with smaller breast tumors have better survival rates than those with large tumors12. SWE as an add-on to B-mode US can measure the elasticity of breast tumors reproducibly and has promising value for predicting the NAC response13–15. When breast cancer cells infiltrate the extracellular matrix (ECM), the surrounding part of the tumor becomes stiffer, leading to breast cancer progression and chemotherapeutic resistance15. It is known that residual tumors are related to prognosis risk, and research on the specific impact of the size and stiffness of breast cancer before and after NAC on prognosis is still limited.
The expression statuses of the estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (Her-2) in breast cancer determine the efficacy of therapy and are the primary prognosis predictors16. An increased probability of achieving pCR is correlated with ER, PR and Her-217. It has been reported that breast tumor size after NAC is significantly correlated with PR before and after NAC18. Moreover, ER positivity, PR positivity and Her-2 negativity are significantly correlated with increased tumor stiffness19. To date, few reports have documented the use of US, CEUS and SWE for accurate imaging evaluation of the NAC response in patients with breast cancer. Our study focused on the accuracy of imaging modalities in determining pCR. Therefore, we aimed to evaluate the predictive value of multiple tumor-associated covariates, including patient age, tumor histology, receptor status, tumor size on US and CEUS, and the maximum stiffness on SWE, for achieving pCR. These results may help advance the understanding of the treatment effect and might eventually lead to better NAC strategies for invasive breast cancer.

Methods

Patient selection, ethical approval and consent to participate
A total of 111 female patients with stage II-IV primary breast cancer whose histopathological characteristics and molecular subtypes of the tumors were confirmed via ultrasound-guided core needle biopsies received NAC at our hospital from November 2018 to July 2023. A flow diagram of the study patient inclusion process is shown in Fig. 1. This study was approved by the Ethics Committee of Zhong Da Hospital of Southeast University (2017ZDKYSB160) on May 17, 2017. All patients provided written informed consent. All methods were performed in accordance with the relevant guidelines and regulations. The authors were not provided with information that could identify individual participants during data collection. We compared the characteristics of multiparametric US imaging between patients who achieved pCR after NAC and those who did not achieve pCR.

US, CEUS and SWE examinations
Conventional breast ultrasonography was performed using a LOGIQ E9 system (GE Healthcare, USA) equipped with an ML6-15 linear transducer. The ML6-15 probe was used to localize the primary breast cancer and obtain the section showing the maximum tumor diameter. Subsequently, a higher-resolution L8-18i probe was employed for detailed scanning to evaluate lesion margins, internal echotexture, and microcalcifications. Any suspicious abnormality detected at the previous tumor site was considered a residual lesion, and the maximum diameter of the hypoechoic area on grayscale ultrasound was recorded as the residual tumor size.
CEUS was performed using the same LOGIQ E9 system with a 9 L linear transducer (5–13 MHz). After selecting the grayscale image displaying the largest cross-section of the tumor, 4.8 ml of the US contrast agent SonoVue (Bracco, Italy) was injected via the antecubital vein, followed immediately by a 5 ml saline flush. Real-time dual-display imaging was used to continuously compare contrast enhanced and grayscale images until the enhancement of the lesion had completely disappeared. Dynamic images were stored for 120–240 s after contrast injection. The maximum lesion diameter, enhancement intensity (hyperenhancement, isoenhancement, hypoenhancement, or nonenhancement), contrast agent inflow time (fast, synchronous, or delayed), and post-enhancement range (increased, unchanged, shrinkage or nonenhanced) in breast cancer lesions were assessed. The presence of filling defects or other perfusion abnormalities was also evaluated.
SWE was performed using an Aixplorer ultrasound system (SuperSonic Imagine, France) equipped with an L15-4 linear transducer (4 − 15 MHz). After identifying the mass with the largest diameter, the mode was switched to elastography mode without applying external compression. Tissue elasticity was displayed as a color-coded map ranging from blue (soft) to red (hard). The sampling frame was moved to the area with highest stiffness, and the maximum Young’s modulus (Emax, range: 0–300 kPa) was measured within the region of interest (ROI) placed on the stiffest part of the lesion. The ROI settings for SWE were determined in a lesion-specific manner according to imaging characteristics before and after NAC. Prior to NAC, some lesions were relatively large and could not be fully delineated using a single free tracing to measure the maximum elasticity modulus (Emax); therefore, for lesions exhibiting heterogeneous color distribution on the elasticity map, a circular ROI was placed over the darkest area, corresponding to the region of highest stiffness. In contrast, for lesions achieving pCR after NAC and displaying a homogeneous blue elasticity map, where the stiffest region could not be reliably identified, a free ROI encompassing the entire lesion was delineated to obtain the Emax value. Three SWE images were acquired at the same lesion, and repeated measurements were performed to confirm and consistently identify the stiffest region within the lesion as the same ROI. Among these acquisitions, the image with optimal signal stability and minimal artifacts was selected for subsequent analysis.
During the initial ultrasound examination, the measurement site for tumor maximum diameter, the distance from the lesion to the nipple and the probe orientation were documented, and corresponding body landmarks were marked on the sonographic images. All subsequent ultrasound assessments were performed strictly at the same marked location using the identical probe orientation. All three US examinations were conducted by two specialized breast sonographers, each with more than 20 years of experience in breast ultrasound diagnosis and over 24,000 independently completed breast ultrasound examinations. Given that patients undergoing NAC typically complete 6–8 treatment cycles, it was not feasible in routine clinical practice to arrange for two sonographers to perform simultaneous and independent real-time acquisitions of CEUS and SWE data for the same patient at each follow-up time point. Therefore, interobserver reproducibility analysis in this study was restricted to baseline measurements obtained prior to NAC. Baseline imaging data from 60 patients were used to evaluate interobserver agreement for two key parameters: the CEUS-measured maximum tumor diameter and SWE-derived Emax. Specifically, one sonographer performed the baseline examination and archived representative images in a standard manner, while the second sonographer independently re-measured both parameters in a blinded manner, based solely on the stored images. To assess measurement reproducibility, interobserver agreement was analyzed using a two-way random-effects model with intraclass correlation coefficients (ICC [2,1]).
Study participants were subjected to US, CEUS and SWE at two time points: within 1 week before starting NAC (DUS0, DCEUS0, Emax0) and after completing NAC approximately 1 day before surgery (DUSend, DCEUSend, Emaxend). Changes in the maximum diameter of the lesion (ΔDUS, ΔDCEUS) and stiffness (ΔEmax) after NAC were compared. US, CEUS or SWE changes were calculated using the following formula20:

Pathological examination and response evaluation
Histopathological evaluation of surgical samples was performed by breast pathologists with more than 15 years of experience using the Miller-Payne grading (MPG) scale for pathological response: Grade 1 – Grade 521. The Grade 5 measurement indicated no identifiable malignant cells or remaining ductal carcinoma in situ. The measurements of Grade 1 – Grade 4 tumors were defined as follows: no reduction in total tumor cellularity; up to 30% loss of tumor cells; 30%−90% reduction in tumor cells; and > 90% disappearance of tumor cells, respectively. When the degree of regression of microscopic breast cancer tumor cell numbers before NAC was compared with the post-NAC findings, Grade 5 was defined as the pCR group, while Grades 1–4 were defined as the non-pCR group.

Results

Patient characteristics and pathological treatment response

Statistical analysis
Based on the interobserver agreement analysis, both the maximum tumor diameter measured by contrast-enhanced ultrasound (CEUS) and the maximum elasticity modulus (Emax) measured by shear wave elastography (SWE) demonstrated good interobserver consistency. The corresponding ICCs with 95% confidence intervals (CI) were 0.931 (0.887–0.958) and 0.911 (0.800–0.955), respectively. The data were analyzed with SPSS version 20 statistics software (IBM Corp., Armonk, NY, USA). The results were visualized with appropriate figures, including the ROC curve and the histogram graph, using R (version 4.41) and GraphPad Prism version 9.5.0 (GraphPad Software, San Diego, California, USA) respectively. Normal distribution of the data was assessed via Shapiro – Wilk analysis. A Levene test was used to evaluate the homogeneity of variance. Continuous variables were presented as mean and standard deviation and were assessed using Wilcoxon rank-sum test. Categorical variables were analyzed using χ2 or Fisher exact test. DUS, DCEUS and Emax prior to and after NAC were compared by Wilcoxon signed-rank test. Multivariate logistic regression model was constructed to predict risk factors including all parameters with significant differences in the univariate logistic regression models. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were obtained for CEUS and SWE to evaluate NAC efficacy. All statistical tests were performed at the 2-sided, α = 0.05 significance level.
The current study included available data from 60 female patients with breast cancer treated with NAC and surgery. According to the postoperative pathological data, 28
patients achieved pCR (46.7%), and 32 patients did not achieve pCR(53.3%). The mean age (± standard deviation) of the 60 patients was 52.0 ± 10.4 years (range, 24–71 years). All 60 patients had invasive breast carcinoma. Immunohistochemical (IHC) biomarkers and subtypes were determined. Table 1 shows that there were significant differences between the pCR group and the non-pCR group in terms of estrogen receptor, progesterone receptor and Her-2 expression (p < 0.05). The pCR rate was highest in patients with Her-2 + breast cancer (53.6%), followed by those with luminal B breast cancer (25%), and triple-negative breast cancer (21.4%).

Tumor size in US and CEUS and Emax in SWE evaluation
Before NAC, all lesions displayed pronounced enhancement on CEUS. In terms of elasticity distribution, most tumors presented peripheral high stiffness, whereas a minority exhibited high stiffness both internally and peripherally. There were no statistically significant differences in DUS, DCEUS or Emax between the pCR and non-pCR groups before NAC, as shown in Fig. 2 (p = 0.439, 0.864, 0.356, respectively). In the pCR group after NAC, CEUS demonstrated no evident contrast
agent perfusion within the target lesions, or only minimal and localized enhancement.
Meanwhile, SWE showed that in most lesions, the Emax value of the stiffest region exhibited no significant difference from that of the surrounding normal breast tissue, indicating that their stiffness levels had become comparable, (Fig. 3). Compared with those in the non-pCR group, a smaller maximum diameter on CEUS, (0.28 ± 0.87 cm vs. 1.97 ± 1.34 cm, p < 0.001) and lower tumor stiffness on SWE, (34.81 ± 22.47 kPa vs. 95.42 ± 83.50 kPa, p < 0.001) were associated with pCR, (Figs. 2 and 4). The results indicated that the changes in DUS, DCEUS and Emax before and after NAC were significantly different between the two groups, while in both cases the tumor sizes and stiffness showed a clearly significant decreasing trend at the two time points, as shown in Table 2.

(Arrows indicate that the upper limit of the 95% confidence interval extends beyond the displayed range of the x-axis.)
After NAC, the percentage of reduction in diameter observed via US (54% vs. 37%, = 0.019), the percentage of reduction in diameter observed via CEUS (93% vs. 48%, p < 0.001), and the percentage of reduction in tumor stiffness (76% vs. 51%, p = 0.004) via SWE were found to be more significant in the pCR group, as shown in Table 3. Our research evaluated the correlations among ΔDUS%, ΔDCEUS%, and ΔEmax% in patients with breast cancer and the prognostic features of heterogeneity. Breast cancers with different IHC biomarkers respond differently to NAC, resulting in significant differences in ΔDUS%, ΔDCEUS% and ΔEmax%, as shown in Table 4. For ER+, PR + and Her-2 + patients, there was a discrepancy in the change in the invasive cancer maximum diameter via CEUS (ΔDCEUS%) between the pCR group and the non-pCR group. The ΔDUS% was significantly different in Her-2 + breast cancer patients, and the ΔDCEUS% in the pCR group was significantly greater than that in the non-pCR group across all the IHC biomarkers. ER- breast cancer, PR- breast cancer and Her-2- breast cancer were significantly different in terms of breast pCR, with a ΔEmax%. Significant changes in ΔDCEUS% and ΔEmax% occurred in triple-negative patients in two groups following NAC (p = 0.002 and 0.003, respectively).
In the univariate regression analysis, factors with a significant predictive value were ΔDCEUS%, ΔDCEUS, DCEUSend, ΔEmax%, Emaxend, ΔDUS%, ER, PR and HER-2 (Fig. 5). In our study, variables with significant differences in the univariate analysis were included in the multivariate logistic regression analysis, and a backward stepwise regression approach was subsequently applied to systematically screen and optimize the included variables. The results demonstrated that ΔDCEUS% and ΔEmax% were significant predictors of pCR (Fig. 5). For ΔEmax% and ΔDCEUS% alone, the areas under the ROC curve were 0.722 and 0.871, respectively. When the two parameters were combined, the area under the ROC curve was 0.906 (Fig. 6).

Discussion
In this study, we used US, CEUS and SWE to explore the changes in tumor size and stiffness after NAC treatment in patients with breast cancer, and we found that ΔEmax% and ΔDCEUS% were independently correlated with pCR. Furthermore, the combination of these two parameters could result in a greater AUC value for predicting pCR after NAC than the use of the two parameters alone. These findings suggest that CEUS and SWE, which are two noninvasive and repeatable ultrasound techniques, hold potiential clinical value for assessing the pathological response to NAC in patients with breast cancer.
US plays an important role in predicting the results of NAC in breast cancer patients. However, the measurement of the tumor size via US is only moderately correlated with residual pathological tumor size after NAC22. In our study, 17 pCR patients misdiagnosed as non-pCR by US presented fibrosis and chronic inflammatory cell infiltration, but no residual malignant lesions remained. These tumor sizes were overestimated because US cannot distinguish the degeneration and necrosis or fibrous scarring from the residual cancer after NAC. Thus, judging the size of breast tumors only by the echo interface of conventional US can result in inaccuracies during NAC23. Although ultrasound-based deep learning has been widely applied for the assessment of NAC in recent studies, it also has a similar predictive value as conventional ultrasound for pCR in these stages24,25. Tumor blood vessels may become thinner or even occluded, and color doppler cannot detect low-velocity blood flow during NAC, thereby limiting the ability of this method to evaluate treatment re-sponse26. Therefore, more ultrasonic imaging features are needed to accurately evalu-ate the tumor response after NAC.
Angiogenesis is a critical factor in the development of breast cancer. Increased levels of angiogenesis are associated with reduced survival in breast cancer patients27. CEUS mainly observes microperfusion inside the tumor, which reveals enhancement patterns and tumor sizes at the initial and final NAC therapy stages. In this study, all tumors in the pCR group were hyperenhanced before NAC, but no or few contrast agents entered the lesions after NAC. Both US and CEUS failed to detect breast cancer lesions in some pCR patients after NAC, and the location of the region of interest could not be accurately determined. Therefore, CEUS quantitative parameters, such as the maximum intensity of the time‒intensity curve during bolus transit (PEAK) and time to peak (TTP), are not entirely appropriate for assessing NAC. Although previous studies have demonstrated that parameters such as peak enhancement (PE), TTP, and mean transit time (mTT) have value in assessing the response to NAC in breast cancer, the quantitative evaluation of these parameters remains subject to certain limitations28. This observation is consistent with the conclusions drawn from the present study. Due to the diversity and unevenness of retraction patterns of breast tumors after NAC, and to minimize the influence of subjective factors and individual differences on the interpretation of CEUS parameters, tumor size determined by CEUS is a natural and commonly used clinical index to analyze the efficacy of NAC in patients with breast cancer29. ΔDCEUS% in the pCR group significantly differed from that in the non-pCR group. A possible explanation could be the difference in the decrease in the concentration of vascular growth factor in the tumors between the two groups. Changes in the size of breast cancer tumor via CEUS are related to blood perfusion and pathological features after NAC20.
SWE has been found to be significantly correlated with tumor growth, thus indicating that it could be another valuable tool for predicting pCR in patients with invasive breast cancer30,31. In our study, a few breast cancer lesions exhibited high stiffness both within and around the tumor, whereas the majority showed the highest stiffness at the tumor periphery. Changes in the stiffness of breast cancer after NAC can accurately predict pCR, while high levels of tumor stiffness have been shown to be correlated with tumor progression and resistance to NAC32,33. Research has shown that different mean elasticities are obtained for the same type of breast cancer with different ultrasound sections, so the maximum elasticity is one of the most reproducible SWE features14,34. We found that the ΔEmax% of lesions in the pCR group was significantly greater than that in the non-pCR group after NAC, indicating that the degree of collagen crosslinking, caveolin, fibronectin, and lysyl oxidase changed to different extents after NAC15,35,36.
Previous studies have shown that molecular biomarkers can provide valuable insights into the heterogeneity of breast cancer. The expression levels of ER, PR and Her-2 by pathologic confirmation determine the optimal drug combination regimen of neoadjuvant chemotherapy, thus eliminating the risk of ineffective treatment, and underscoring the leading role pathological analyses have in achieving a complete response18,37,38. Her-2 overexpression, a negative estrogen receptor status, and a triple-negative receptor status are associated with an increased probability of achieving pCR17,39. Our study confirmed that negativity for ER and PR receptors is significantly correlated with high chemosensitivity and can result in a smaller tumor size. However, TNBC patients with high histological malignancy often present with benign lesion features, and tumor stiffness seems to be more helpful for differentiating fibroadenomas from TNBC40. For ER-, PR- and Her-2- breast cancer, there were significantly greater values of ΔDCEUS% and ΔEmax% in the pCR group than in the non-pCR group. This phenomenon may indicate a relationship among immunohistochemical factors, ΔDCEUS% and ΔEmax%; furthermore, this association, which predicts tumor biological activity, can guide treatment options for breast cancer patients41.
Breast cancer patients with pCR have a better prognosis than non-pCR patients who receive the same regimen42. CEUS and SWE can be used to forecast the efficacy of neoadjuvant chemotherapy in patients with breast cancer30,43. When CEUS cannot distinguish the degeneration and necrosis or fibrous scars from the residual cancer after NAC, breast tumors in the non-pCR group may have higher collagen and hyaluronic acid contents in the extracellular matrix and higher interstitial fluid pressures. Higher interstitial fluid pressures cause blood vessel collapse, tissue hypoperfusion and drug delivery interference because of fast-growing cancer, resulting in a reduction in lesion enhancement and a relatively high stiffness44,45. In contrast, when the breast tumor is small or deep, the ΔEmax% may not change significantly after NAC, and CEUS could help reveal the real change in the size of the residual extent of the lesion. Our multivariate analysis revealed that ΔDCEUS% combined with ΔEmax% was independently associated with pCR after NAC. The combination of ΔDCEUS% and ΔEmax% provided significantly greater diagnostic performances than each parameter alone. Effective neoadjuvant chemotherapy can trigger the apoptosis of immature vascular endothelial cells, slow their proliferation, reduce the density of cancer cells, thus leading to tumor shrinkage or disappearance46–48. We observed differences in the likelihood of NAC success, and the combined application of ΔDCEUS% and ΔEmax% as effective indicators may demonstrate more favourable performance in predicting the pathological response to NAC among patients with invasive breast cancer. These findings are in line with previous studies, which reported that patients achieving pCR tended to have markedly smaller tumor size on CEUS, lower Emax value on SWE and a higher frequency of hormone receptor negativity compared with those in the non-pCR group49,50.
There are several limitations of our study. First, the small sample size reduces the stability of the regression model, and more clinical samples are needed to validate the proposed method. The immunohistochemical subtypes were limited: there was only one luminal A subtype. As a result, we could not analyze the correlation among the immunohistochemical subtype, CEUS and SWE parameters. Second, owing to intratumor heterogeneity and diverse molecular and phenotypical profiles, ultrasound imaging has not been used for the early assessment of pathological complete response to NAC. Third, breast cancer lesions can be selected in only one plane at a time via contrast-enhanced ultrasound; thus, our study investigated only cases in which a breast cancer lesion retreated but remain one lesion, rather than changing into multiple lesions after neoadjuvant chemotherapy. Finally, a limitation of our reproducibility assessment is that the interobserver comparison was based on retrospective re-measurements of static stored images rather than independent real-time scans, which may not fully capture variability introduced during live image acquisition.

Conclusion
The combination of CEUS and SWE is a feasible and efficient approach for evaluating response to NAC in patients with invasive breast cancer. Good treatment responses were associated with higher ΔDCEUS% and ΔEmax%. The combined use of these imaging techniques enables reliable prediction of complete pathological response and offers a highly accurate method for evaluating treatment-induced changes in breast cancer.